A brand voice rarely survives the border into the next language intact. The text comes back correct, every word in place, and yet the brand suddenly sounds like somebody else. The reason is almost never the quality of the translation. It is that nobody defined what about the voice was supposed to travel in the first place.
- A brand voice is transferable, a tone made of example sentences is not.
- What gets translated is the intent, not the sentence structure.
- Without a decision-maker in the target language, grammar decides.
- A voice only becomes testable once it produces a no.
Table of contents
Why a brand voice disappears in translation
The trigger for taking the subject seriously is usually commercial. In its study Can’t Read, Won’t Buy, CSA Research found that 76 per cent of shoppers prefer product information in their own language and a quarter will not buy at all without it. In many companies the conclusion drawn from that is to translate everything. What does not follow from it is an answer to the question of how the brand should sound in those languages.
What I keep seeing in these projects is a document with three adjectives and five sample sentences. In the original language it works, because everyone involved has the same music in their ear. The moment the same ambition moves into English or Chinese, only the example survives, and an example is the weakest possible form of an instruction. A brand voice described only in sample sentences cannot travel.
The second cause is the shape of the assignment. Translation gets bought as a service, not as a decision. The supplier receives a text and a deadline but no position. So they deliver the version least open to attack, and that is reliably the most neutral one. Neutral is the state in which a brand voice stops existing.
What transfers and what does not
The useful dividing line runs between intent and means. The intent of a brand is translatable, its linguistic means rarely are. Mix the two in one document and you force the target language into the melody of the source, then wonder why the result reads stiff. Drawing the line costs effort once and saves it in every further language.
What has to stay
- Distance. How close the brand steps towards the reader.
- Demand. How much prior knowledge it assumes.
- Pace. Whether it asserts or derives.
- Humour. Whether it allows any, and exactly where.
What is allowed to change
Sentence length, word choice, imagery and forms of address belong to the target language, not to head office. In Chinese a short sentence carries a different authority than it does in German, and English tolerates a directness that reads as presumptuous elsewhere. Fix that layer centrally and you are not exporting a brand voice, you are exporting a grammar.
The same holds for word lists. Banned terms are useful in the source language and often meaningless in the target one, because the register you meant hangs on a different word there. A brand voice steered through a list of prohibitions loses substance with every language added, until all that remains is a collection of terms nobody may use and no picture at all of how the brand actually speaks.
The fault sits in the brief to the translator
I believed for a long time that the problem could be solved with better translators. It gets solved with better assignments. A brief consisting of a source text and a deadline produces a copy. A brief that names the intent of the text produces an equivalent. The difference costs two sentences and decides the outcome.
What has worked in practice is a short line above every text: what it should trigger in the reader, and what it explicitly must not be. The second half matters more, because it draws the boundary. The same discipline I consider necessary for internal briefs in an in-house agency applies here unchanged.
Sign-off is the other half of the problem. When a text is approved in a language the approver does not speak, nothing has actually been approved. A brand voice needs one person per language who will defend it when it is questioned, and that person has to be named before the first text, not after the first complaint.
How to make a brand voice testable
The most useful test is always the same one. Can a no be derived from the description? A brand voice that only describes how the brand sounds helps nobody. One that names what it would never do settles arguments. We tend to work with negative pairs here, two versions of the same sentence where one is explicitly wrong. That is more effort than a list of adjectives and it survives the change of language.
The second test is a back-translation pass, though not the literal kind. I ask somebody who has never seen the source to summarise the foreign-language text in their own words, then compare that summary against the original intent. What surfaces is rarely an error. It is a shift. And shifts are precisely what dissolves a brand voice over the course of a few years.
What a staged build looks like
Anyone trying to make a brand voice hold across languages should not start with a handbook but with a single language. The order that has proved itself is always the same.
- First. Separate the intent of the voice from its linguistic means.
- Second. Name one responsible person for every language.
- Third. Work out the negative pairs in exactly one target language.
- Fourth. Only then roll the result out to the remaining languages.
The most common mistake is starting at stage four. A handbook gets written for all markets at once, tested in none of them, and is therefore read as equally non-binding everywhere. A document that has settled one real conflict in one language gets taken seriously in the second.
Measurement misleads here with some regularity too. Translation quality gets measured by error rates, because errors are countable. That does not make a brand voice better, only cleaner. The more useful question is whether a reader in the target language recognises the same brand, and that can only be answered qualitatively.
What I took away from it
My biggest misjudgment was treating this as a question of language competence. It is a question of clarity in the original. A text whose intent is blurred in the source gets blurrier in every further language, because each translation makes a decision nobody made before. Anyone who wants the transfer to improve has to start with the source text.
What surprised me is how strongly the detour feeds back into the original. Once a brand voice is described well enough to be reproducible in a foreign language, it becomes more precise in its own. How I sort this into the wider picture is under brand leadership. More on this in the Creation & Storytelling section.